Growth Rate and Cell Size Modulate the Synthesis of, and Requirement for, G₁-Phase Cyclins at Start

Abstract

In Saccharomyces cerevisiae, commitment to cell cycle progression occurs at Start. Progression past Start requires cell growth and protein synthesis, a minimum cell size, and G₁-phase cyclins. We examined the relationships among these factors. Rapidly growing cells expressed, and required, dramatically more Cln protein than did slowly growing cells. To clarify the role of cell size, we expressed defined amounts of CLN mRNA in cells of different sizes. When Cln was expressed at nearly physiological levels, a critical threshold of Cln expression was required for cell cycle progression, and this critical threshold varied with both cell size and growth rate: as cells grew larger, they needed less CLN mRNA, but as cells grew faster, they needed more Cln protein. At least in part, large cells had a reduced requirement for CLN mRNA because large cells generated more Cln protein per unit of mRNA than did small cells. When Cln was overexpressed, it was capable of promoting Start rapidly, regardless of cell size or growth rate. In summary, the amount of Cln required for Start depends dramatically on both cell size and growth rate. Large cells generate more Cln1 or Cln2 protein for a given amount of CLN mRNA, suggesting the existence of a novel posttranscriptional size control mechanism.

In most eukaryotes, cells become committed to a round of cell division at an event in G₁ phase. This event is called “Start” in the yeasts (14), and the “restriction point” in mammalian cells (2, 42). In yeast, passage through Start seems to require growth to a minimum cell size. For example, a wild-type W303 haploid strain in rich medium with glucose as the carbon source progresses past Start at a cell size of roughly 35 fl. While growth to a minimum cell size is necessary for commitment to cell division, it is not sufficient—active protein synthesis is also required (18, 28, 42). This suggests that unstable proteins must accumulate for cell cycle progression. It is widely believed that these proteins are G₁-phase cyclins (reviewed in references 2, 3, 11, 21, 27, and 34).

In Saccharomyces cerevisiae, it is well established that Start is dependent upon the G₁-phase cyclin Cln3 and its cyclin-dependent kinase (Cdk) Cdc28 (reviewed in references 3, 11, 21, 27, and 34). The amount of Cln3-Cdc28 is relatively constant throughout the cell cycle (37). However, just prior to Start, as cells approach the required minimum size, Cln3-Cdc28 kinase somehow activates two transcription factors, called SBF and MBF, and induces the transcription of about 200 genes (12, 33). Among these SBF and MBF targets are two more G₁-phase cyclins, Cln1 and Cln2, and two S-phase cyclins, Clb5 and Clb6 (12, 33). These cyclins also form protein kinase complexes with Cdc28, which phosphorylate various substrates, and ultimately push cells into S phase. One particularly well-studied event is the phosphorylation of Sic1, a Cdk inhibitor, by the Cln1-Cdc28 and Cln2-Cdc28 complexes (19, 30, 31, 36). Once Sic1 has been phosphorylated, it is ubiquitinated and degraded, and this allows activation of the Clb5-Cdc28 and Clb6-Cdc28 complexes, and these directly activate DNA replication (19, 30, 31, 36). Analogous events also occur in mammalian cells, where a G₁-phase cyclin, in this case cyclin D, forms a complex with Cdk4 and activates E2F and related transcription factors (reviewed in reference 8). In this case the activation occurs, at least in part, through phosphorylation of the retinoblastoma protein, which otherwise inhibits E2F activity. Thus, Cln3 works through several downstream effectors, including two critical downstream G₁-phase cyclins, to promote progression past Start (6, 7, 37).

As G₁-phase cells grow in size toward Start, the abundance of the Cln1 and Cln2 mRNAs and proteins increases (7, 37). This suggests that protein synthetic and cell size requirements may in part reflect the need for a critical amount of the Cln1 and Cln2 proteins. In fact, in both yeast and mammalian cells it is widely believed that a critical amount of G₁-phase cyclins must accumulate to induce cell cycle progression (reviewed in references 2, 3, 11, 21, 27, and 34). However, this hypothesis has not yet been analyzed quantitatively.

A closely related issue is the relationship between growth rate and G₁-phase cyclin abundance. Mass doubling times for yeast range from 90 to at least 600 min under different growth conditions. Because CLN mRNAs and proteins are unstable, with half-lives in the vicinity of 5 to 10 min, their abundance is in a constant equilibrium with their rate of synthesis (4, 17, 29, 38, 41). This is advantageous from a regulatory point of view, since interruption of protein synthesis immediately aborts progress toward Start. However, it poses a serious hypothetical problem for slowly growing cells—if Cln half-lives remain short even when the rate of protein synthesis is low, then the levels of Cln protein will be very low at low growth rates (Fig. 1). Because Cln1 and Cln2 are unstable at both the mRNA and protein levels and because their transcription depends on Cln3, which is likewise unstable at both the mRNA and protein levels, one could imagine that a small decrease in protein synthesis rates could have an enormous multiplicative effect on the abundance of Cln1 and Cln2 (Fig. 1) (4, 17, 29, 38, 41). This line of thinking suggests that rapidly growing cells might have large amounts of Cln while slowly growing cells might have smaller amounts. Indeed, Heideman and coworkers have clearly demonstrated that the CLN3 mRNA is abundant when cells are grown on glucose and much less abundant under any other growth condition (22, 23, 25, 26, 40). Tokiwa et al. have suggested that the cyclic AMP pathway partly compensates for the large amounts of Cln3 in glucose cultures by reducing the abundance of CLN1 transcript under these conditions (35).

FIG. 1.

The slow-growth problem. Unstable proteins achieve equilibrium levels proportional to their rate of synthesis. Thus, slowly growing cells can never accumulate the same amount of Cln protein as rapidly growing cells. Here, the abundance of a stable protein is compared with the abundance of an unstable protein at two different protein synthesis rates, rapid and slow. The dotted line represents a hypothetical critical threshold amount.

Thus, on the one hand, Cln levels increase as cells grow larger but on the other hand, there are theoretical and some experimental reasons to believe that slowly growing cells have less Cln than do rapidly growing cells. How, then, does a slowly growing cell ever attain sufficient Cln protein to get past Start? Does the hypothetical dependence of Cln1 and Cln2 protein abundance on the growth rate actually occur, and if it does, is there a compensatory mechanism to allow Start when the rate of protein synthesis (and Cln abundance) is low? Does the amount of Cln have to rise to a critical threshold to promote Start, or can a small amount of Cln acting over a long period of time suffice?

We report here that Cln1 and Cln2 protein levels are strongly regulated by the growth rate. Slowly growing cells express dramatically less Cln1 and Cln2 protein than do rapidly growing cells. However, they compensate by somehow requiring less Cln for Start. A threshold level of Cln seems to be necessary to promote cell cycle progression, but this threshold level varies tremendously, depending on the conditions—the threshold level is much higher for rapidly growing cells and for smaller cells. Finally, Cln protein expression is cell size dependent, in that large cells appear to contain much more Cln protein per unit of CLN mRNA than do small cells, and this suggests the existence of a (translation-based?) size control mechanism independent of the CLN3/SBF/MBF pathway. Elucidation of the mechanisms responsible for these surprising observations will be an important step in uncovering the molecular details that coordinate cell growth with division.

MATERIALS AND METHODS

Strains and media.

The strains used in this work were derived from strain W303 (Table 1). Yeast cultures were grown in YEP-based medium (20.0 g of Difco Bacto Peptone and 10.0 g of Difco Bacto Yeast Extract dissolved in 900 ml of water and autoclaved) or YNB-based medium (6.7 g of Difco Bacto Yeast Nitrogen Base lacking amino acids and ammonium sulfate added to 900 ml of water and autoclaved; required amino acids were supplemented at 50 mg/liter, except for l-tryptophan [80 mg/liter], adenine sulfate [32 mg/liter], and p-aminobenzoic acid [5 mg/liter]). After autoclaving, sterilely filtered carbon sources were added to a final concentration of 2% (glucose, sucrose, or raffinose) or 3% (ethanol and glycerol).

TABLE 1.

Strains used in this study

Strain	Genotype
W303	MATaade2 can1 his3 leu2 trp1 ura3
GT104	MATaade2 can1 his3 leu2 trp1 ura3 CLN1-HA3
GT108	MATaade2 can1 his3 leu2 trp1 ura3 CLN3-HA3
BS111	MATα ade2 can1 ura3 leu2::GAL-CLN1-HA3 cln2::TRP1 cln3::HIS3 gal1 gal10::hisg
GT123-6b	MATaade2 can1 his3 leu2 trp1 ura3 rpb1-1 CLN1-HA3
GT124-6a	MATaade2 can1 his3 leu2 trp1 ura3 rpb1-1 CLN2-HA3
GT125-9c	MATaade2 can1 his3 leu2 trp1 ura3 rpb1-1 CLN3-HA3
GT130-14	MATaade2 can1 his3 trp1 ura3 reg1 gal1 leu2::GAL-CLN1-HA3
GT141-1	MATaade2 can1 his3 trp1 ura3 reg1 gal1 leu2::GAL-CLN2-HA3
MT263	MATaade2 can1 his3 leu2 trp1 ura3 CLN2-HA3

Expression of a controlled amount of CLN1 mRNA from the GAL promoter.

To create a strain in which the GAL promoter could be induced by very low concentrations of galactose, we deleted the promoters of the divergently transcribed GAL1 and GAL10 genes in a conditional CLN strain (cln1 cln2 cln3 GAL-CLN1), yielding gal1 gal10 mutants (BS111). To express a controlled amount of CLN1 mRNA, strain BS111 was grown in YEP-1% raffinose-1% galactose to mid-log phase, washed thoroughly with YEP-1% raffinose, and resuspended in YEP-1% raffinose for several hours to ensure that the GAL-CLN1 construct was shut off. For Northern analysis, cultures were grown to 1 × 10⁷ to 3 × 10⁷ cells per ml. To confirm that the cell concentration did not have an effect on the ability of galactose to induce the GAL-CLN1 gene at low doses of galactose, experiments were repeated at cell concentrations of ∼10⁵/ml. To express defined amounts of CLN1 in different-sized cell populations, centrifugal elutriation was used to obtain small unbudded cells. These small unbudded cells (∼10 to 15 fl) were resuspended in fresh YEP-1% raffinose and split into eight fractions at a concentration of 5 × 10⁶ to 10 × 10⁶ cells per ml. In each fraction, different amounts of galactose were added (final percentages: 0, 0.001, 0.003, 0.01, 0.03, 0.1, 0.3, and 1%) and cells were incubated at 30°C. Samples were taken at regular intervals, and cell size and progression past Start were measured as a function of the percentage of budded cells. To produce larger unbudded G₁-phase cells, initial elutriated fractions were resuspended in fresh YEP-1% raffinose and incubated at 30°C until the desired size was achieved.

Preparation of RNA and Northern analysis.

Yeast cultures were grown to mid-log phase (1 × 10⁷ to 3 × 10⁷ cells/ml). Cultures were chilled rapidly by adding an equal volume of ice to the medium. The cells were pelleted by centrifugation at 4°C and washed in ice-cold water. Cell pellets were frozen at −80 °C. Pellets were resuspended in 250 μl of LETS buffer (100 mM LiCl, 10 mM EDTA, 10 mM Tris-HCl [pH 7.4], 0.2% sodium dodecyl sulfate [SDS]). Subsequently, 300 μl of LETS-equilibrated phenol and an equal volume of 450-nm acid-washed glass beads were added. The cell suspensions were vortexed at maximum speed for 30 s, and then an additional 200 μl of LETS was added. The cell suspensions were vortexed briefly and then centrifuged for 5 min at 14,000 rpm (16,000 × g). The upper aqueous phase was removed and extracted twice with phenol-chloroform. RNA was precipitated by adding 1/10 volume of 5 M LiCl and 2.5 volumes of ice-cold ethanol and incubated for 1 to 12 h at −20 °C. After precipitation, RNA was recovered by centrifugation for 15 min at 14,000 rpm (16,000 × g), followed by a wash with 70 to 80% ethanol. RNA pellets were air dried at room temperature and resuspended in 50 to 100 μl of diethyl pyrocarbonate (DEPC)-treated water. Size separation of RNA was performed with 1.0% denaturing agarose gels containing 6.6% formaldehyde and l× morpholinepropanesulfonic acid (MOPS). Ten micrograms of RNA was lyophilized in a Savant Speed Vac and resuspended in 5 μl of DEPC-treated water. Subsequently, 17.5 μl of RNA loading buffer (12.5 mM MOPS [pH 7.1], 2.5 mM sodium acetate, 0.25 mM EDTA, 3.1% formaldehyde, 25% formamide, 2% glycerol dye, 4 mg of bromphenol blue per ml, 4 mg of xylene blue per ml, 50 μg of ethidium bromide per ml) was heated at 65°C for 15 min. RNA samples were loaded onto a gel that was prerun at 90 V for 20 min. Gels were run at 45 V for 30 min and then at 90 V for 3 to 5 h. After electrophoresis, gels were soaked in DEPC-treated water with gentle shaking for 45 min and then transferred to Nytran-Plus nylon membranes (Schleicher & Schuell) as recommended by the manufacturer. After transfer, nucleic acids were cross-linked to membranes with UV light (UV Stratalinker 1800) as recommended by the manufacturer. Hybridization of membranes was performed as previously described (37, 38) with Church hybridization buffer (7% [wt/vol] SDS, 0.1% [wt/vol] bovine serum albumin [fraction V; Sigma], 0.l mM EDTA, 0.25 M Na₂HPO₄ [pH 7.2]). Filters were preincubated in hybridization buffer for 30 min at 65°C. Radioactive probes were made with [α-³²P]ATP and a random prime labeling kit from Boehringer Mannheim. Probes were purified on Sephadex G-50 spin columns, denatured by boiling for 5 min, added to prehybridization buffer, and incubated for 12 to 16 h at 65°C. Subsequently, blots were washed once with 2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) for 5 min, twice with 2× SSC-0.1% SDS (preheated to 65°C) at 65°C for 15 min, and finally in 2× SSC for 15 min. Filters were wrapped in plastic wrap and exposed to Kodak XAR film or to Fuji phosphorimager screens for further analysis.

Quantification of Northern data was conducted with the FluorChem 2.0 spot densitometry analysis program (Alpha Innotech). Images captured on film were digitized and analyzed. To ensure linearity of the signal from film exposures, three to seven exposures were analyzed in each case but saturated exposures were avoided. Because of the wide dynamic range in some experiments, the film exposures used for quantitation of some lanes do not necessarily include the exposure shown in the figure. To control for loading, CLN mRNA signals were normalized to the ACT1 mRNA.

Analysis of CLN mRNA and protein stability.

The half-life of CLN mRNAs and proteins was determined at various proliferation rates with both rpb1-1 mutant assays and GAL-promoter shutoff experiments. The following isogenic strains containing integrated hemagglutinin epitope-tagged CLN genes and rpb1-1 were constructed: GT123-6b (CLN1-HA3 rpb1-1), GT124-6a (CLN2-HA3 rpb1-1), and GT125-9c (CLN3-HA3 rpb1-1). The rpb1-1 mutant gene encodes a temperature-sensitive allele of RNA polymerase II, such that transcription ceases abruptly upon a shift from 23 to 36°C. Epitope-tagged rpb1-1 mutant temperature-sensitive strains were grown in the desired medium to mid-log phase at the permissive temperature of 23°C. Cells were immediately shifted to the restrictive temperature of 36°C by the addition of an equal volume of 48°C medium. Cells were immediately placed in a 36°C water bath. Samples were taken prior to the temperature shift and at regular intervals following the shift. Typically, 25 to 50 ml of culture from each time point was collected and used for protein extraction, RNA extraction, fluorescence-activated cell sorter, cell size, cell number, and cell budding analyses.

Immunoprecipitations, histone H1 kinase assays, and immunoblot analysis.

Yeast extracts for immunoprecipitations and immunoblot analysis were prepared as previously described (37, 38). Briefly, yeast pellets were lysed in a mini-beadbeater cell disrupter (Biospec) with 0.5-mm-diameter acid-washed baked zirconia beads in the presence of buffer 3 (0.1% NP-40, 250 mM NaCl, 50 mM NaF, 5 mM EDTA, 50 mM Tris-HCI [pH 7.5]) and proteinase inhibitors (1 mM phenylmethylsulfonyl fluoride, l μg of leupeptin per ml, 1 μg of pepstatin per ml, 0.6 mM dimethylaminopurine, 10 μg of soybean trypsin inhibitor per ml, and 1 μg of tosylsulfonyl phenylalanyl chloromethyl ketone [TPCK] per ml). Cell debris was pelleted by centrifugation at 14,000 rpm (16,000 × g) for 15 min. Protein concentrations were determined with the Bio-Rad dye-binding assay in accordance with the manufacturer's specifications. Immunoprecipitations were carried out by incubating 2 to 6 mg of extract with 0.3 μl of the 12CA5 anti-HA3 antibody on ice for 2 h. Next, 10 μl of protein A-agarose beads (Pharmacia) was added and the extracts were rocked at 4°C for 2 h. Beads were washed four times with buffer 3 and twice with 2× kinase reaction buffer (100 mM Tris [pH 7.5], 20 mM MgCl₂, 2 mM dithiothreitol) with pulse spins of <1,000 × g in a microcentrifuge. Beads were transferred to a fresh microcentrifuge tube and incubated in a heat block at 37°C for 10 min. Subsequently, 5 μl of kinase reaction mixture (2 μl of kinase reaction buffer, 1 μl of [γ-³²P]ATP, 0.5 μl of 2 mg of histone H1 per ml, 1 μl of 3 μM ATP, 0.5 μl of water) was added and tubes were incubated at 37°C for another 10 min. Before the histone H1 kinase assay mixtures were loaded onto polyacrylamide gels, 10 μl of 2× protein sample buffer was added and samples were boiled for 2 min. Small SDS-10% polyacrylamide gel electrophoresis gels were run at 75 to 100 V, dried, and exposed to Kodak XAR film or a phosphorimager screen for further analysis.

For immunoblot (Western) analysis, 50 μg of protein lysates was mixed with an equal volume of 2× protein sample and samples were boiled for 2 min. Samples were loaded onto small SDS-10% polyacrylamide gel electrophoresis gels and run at 75 to 100 V. Protein gels were transferred to nitrocellulose with a semidry transfer apparatus (Millipore) and probed consecutively with primary anti-HA antibody 12CA5 (diluted between 1:2,500 and 1:10,000) and secondary horseradish peroxidase-conjugated sheep anti-mouse antibody (1:20,000; Amersham). Proteins were visualized with the Amersham ECL system or the Pierce Supersignal system in accordance with the manufacturer's specifications.

Quantification of Western data was conducted with the FluorChem 2.0 spot densitometry analysis program (Alpha Innotech). Images captured on film were digitized and analyzed. To ensure linearity of the signal, three to seven exposures were analyzed in each case but saturated exposures were avoided. Because of the wide dynamic range in some experiments, the film exposures used for quantitation of some lanes do not necessarily include the exposure shown in the figure. To control for loading, Cln protein signals were normalized to that of β-tubulin.

Quantification of cell size, percentage of budded cells, and cell cycle distributions.

Cell cycle synchronizations were performed by centrifugal elutriation as previously described (29). Cell cycle synchrony was confirmed by microscopic analysis and flow cytometry. The percentage of budded cells was determined by coding samples and then counting the cells with visible buds in a minimum of 200 cells. The percentage of budded cells was verified in at least two independent experiments. α-Factor resistance assays were conducted essentially as previously described (38). For flow cytometry, yeast cells were harvested, washed, sonicated, and fixed overnight in 70% ethanol at 4°C. Cells were resuspended in 50 mM sodium citrate, washed in the same buffer, sonicated, treated with RNase A (final concentration, 0.25 mg/ml) for 1 h at 50°C, and treated with proteinase K (final concentration, 1 mg/ml) for an additional hour at 50 °C. Before analysis, the yeast cells were stained with propidium iodide at a final concentration of 16 mg/ml. Flow cytometry was performed on yeast cells stained with propidium iodide with a FACScalibur (Becton Dickinson) or Epics XL (Beckman-Coulter) flow cytometer as previously described (5). Analysis of the cell size distribution of yeast strains was done with cultures in mid-log phase. Samples of the cultures were resuspended in 10 ml of Isoton buffer, briefly sonicated, and immediately analyzed with a Coulter Counter Channelyzer ZM or Z2.

RESULTS

Cln protein and Cln-associated kinase activity are modulated by growth rates.

We measured the levels of Cln proteins in asynchronous cultures growing at different rates. Cells grown in YEP medium with 2% glucose, 2% raffinose, or 3% ethanol as the carbon source had a mass doubling time of about 90, 130, or 220 min, respectively. Northern analysis showed that CLN1 and CLN2 mRNAs were present at similar levels under all three growth conditions (Fig. 2A and D), consistent with the previous results of Silljé et al. (32). However, Western analysis showed a marked variation in Cln1 and Cln2 protein levels, with the cells grown in glucose having six to seven times more Cln than the cells growing in ethanol (Fig. 2B and E). The cells grown in raffinose had an intermediate amount of Cln. Similarly, when growth rates were varied with different nitrogen sources (proline, arginine, or ammonium sulfate) in synthetic medium, Cln2 levels were high when growth was fast and low when growth was slow (Fig. 2C).

FIG. 2.

Cln protein levels are modulated by growth rate. (A) Northern analysis. Asynchronous cultures were grown in YEP plus 2% glucose, 2% raffinose, or 3% ethanol. These conditions give a cell division time of about 90, 130, or 220 min, respectively. The relative abundance of each CLN mRNA was assayed by Northern blotting. The ACT1 mRNA was used as a loading control. Quantitation is shown in panel D. (B) Western analysis of HA epitope-tagged Cln1 (strain GT104) or Cln2 (strain MT263) from asynchronous cultures growing in glucose, raffinose, or ethanol. Lysates from an untagged strain are shown as a negative control. β-Tubulin (B-Tub) was used as a loading control. Quantitation is shown in panel E. The Cln1-HA and Cln2-HA strains are the same strains used in panel A. (C) Cln2 protein abundance in cells grown in ammonium sulfate, arginine, or proline. Cells were grown in YNB medium with glucose. The nitrogen sources were ammonium sulfate (YNB-AS), arginine (YNB-ARG), and proline (YNB-PRO). For comparison, Cln2 from YEP-2% glucose is also shown (YPD). (D) Quantification of CLN mRNAs. CLN mRNAs shown in panel A were quantified as described in Materials and Methods and normalized to the ACT1 mRNA loading control. The signal in the ethanol cultures was set to 1. (E) Quantification of Cln proteins. The Cln proteins shown in panel B were quantified as described in Materials and Methods and normalized to the β-tubulin loading control. The signal in the ethanol cultures was set to 1.

To see if Cln-associated kinase activity was similarly affected, cells were grown in YEP with glucose, galactose, or glycerol as the carbon source (with a mass doubling time of about 90, 120, or >300 min, respectively). Epitope-tagged Cln1, Cln2, or Cln3 was immunoprecipitated with an anti-HA antibody, and the amounts of protein and associated histone H1 kinase activity were measured (Fig. 3). In this assay, the exogenous substrate, histone H1, and endogenous substrates, including Cln proteins and other unidentified proteins, are phosphorylated. For all three Cln proteins, Cln protein and the detected kinase activity against these substrates were highest in the glucose-grown cells. Those in galactose-grown cells were intermediate. In terms of histone H1 kinase activity, the galactose-grown cells were similar to glucose-grown cells, but in terms of endogenous Cln-associated substrates, Cln1- and Cln3-associated kinase activity was clearly lower in galactose-grown cells than in glucose-grown cells. This difference was smaller in Cln2 cells. However, the key point is that Cln-associated kinase activity is considerably higher in rapidly growing cells (glucose) than in slowly growing cells (glycerol).

FIG. 3.

Cln-associated kinase activity is modulated by growth rate. Asynchronous cultures containing HA epitope-tagged Cln1 (GT104), Cln2 (MT263), or Cln3 (GT108) were grown in YEP plus 2% glucose, 2% galactose, or 3% glycerol. The cell division time was about 90, 130, or >300 min, respectively. Epitope-tagged Cln proteins were immunoprecipitated from protein lysates, and the Cln-associated kinase activities were measured with histone H1 as the substrate. Western analysis of Cln proteins from these lysates showed that Cln protein levels were correlated with the Cln-associated kinase activities (data not shown).

The experiments shown in Fig. 2 and 3 were conducted with asynchronous cells, but the critical issue is the amount of Cln expressed at Start. Thus, it is possible that the small amounts of Cln in slowly growing cells reflect a long period in early G₁ phase in which there is no Cln, followed by a short period in late G₁ phase in which Cln levels might be comparable to Cln levels in rapidly proliferating cells. Indeed, Silljé et al. have shown in very slowly proliferating cells with a G₁ phase of 580 min that CLN1 and CLN2 mRNAs are not significantly expressed until slightly more than half of G₁ phase has passed (this is not true for CLN3 mRNA, however, which was expressed at relatively constant levels throughout G₁ phase) (32, 37). To address this issue, we used elutriation to fractionate cultures according to cell size and cell cycle position and then measure Cln levels just before Start, at Start, and just after Start.

One such experiment is shown in Fig. 4. Twelve consecutive fractions of increasing cell size were taken and assayed for cell size, percent budding, and Cln2 content. The Start fraction, fraction 5, is so defined because it is the first fraction in which 50% of the cells are budded (Fig. 4B). Extracts from the original, asynchronous ethanol-grown culture, and a comparison glucose-grown culture are also included. There are several noteworthy points. First, it is clear that even peak levels of Cln2 in the ethanol culture are much lower than average Cln2 levels in the glucose culture. Therefore, Cln2 levels at Start in ethanol must have been much lower than Cln2 levels at Start in glucose. This point is further affirmed in Fig. 5 (see below). Second, it is clear that the peak of Cln2 protein actually occurs after Start, at fraction 6 (Fig. 4B). Indeed, since 50% of the cells are already budded by fraction 5, it seems likely that commitment to budding and cell division has already occurred by fraction 4. Thus, these cells seemed to pass through Start while Cln2 abundance was still increasing and had not yet reached its peak. This may contribute to the switch-like, irreversible nature of Start.

FIG. 4.

Slowly growing cells express low levels of Cln2. (A) Asynchronous cells containing HA epitope-tagged Cln2 (MT263) were grown in YEP-3% ethanol to mid-log phase. Cells were harvested by centrifugation at 0°C and elutriated. Twelve fractions were collected on ice. Lane G shows Cln2 protein in asynchronous cells grown in glucose. Lane E shows asynchronous cells grown in ethanol (the starting culture for the elutriation experiment). Cln2 abundance was assayed by Western analysis. (B) Quantification of Cln2 protein. Cln2 protein was quantified as described in Materials and Methods and normalized to the β-tubulin (B-Tub) loading control. The signal in the asynchronous ethanol culture was set to 1. The percentage of budded cells is plotted (white line and right y axis). Flow cytometry indicated that DNA synthesis was occurring by fraction 5.

FIG. 5.

Slowly growing cells require relatively little Cln2 for Start. (A) Cells containing HA epitope-tagged Cln2 (MT263) were grown in YEP plus 3% ethanol (lanes 1 to 3), 2% raffinose (lanes 4 to 6), or 2% sucrose (lanes 7 to 9). The cell division time was about 220, 130, or 100 min, respectively. Cultures were grown to mid-log phase, harvested by centrifugation at 0°C, and elutriated. Fractions were collected on ice. Cln2 abundance was assayed by Western blotting. Budding, flow cytometry, and α-factor execution assays were used to confirm synchrony and cell cycle position. The percentage of budded cells in each fraction is given. (B) Quantification of Cln2 protein. Cln2 protein was quantified and normalized to the β-tubulin (B-Tub) loading control. The signal in the first ethanol fraction (lane 1) was set to 1. Note that in sucrose lane 9, fewer than half of the cells have passed Start yet the cells contain much more Cln2 than in ethanol lane 3 or raffinose lane 6, where a majority of the cells have passed Start.

Similar elutriation experiments were then done with cells grown in YEP plus 2% sucrose, 2% raffinose, or 3% ethanol, with doubling times of about 100, 130, or 220 min, respectively. In these cases, selected fractions around Start were chosen for analysis by Western blotting. Synchrony and cell cycle position were determined by the percentage of budded cells, cytometry, and α-factor resistance assays. As shown in Fig. 5, pre-Start cells grown in sucrose (lanes 7 to 9) had far more Cln2 than did cells grown in ethanol (lane 3) or raffinose (lanes 5 and 6) at Start (Fig. 5A and B). The pre-Start sucrose-grown cells had not yet become committed to the cell cycle (as shown by their sensitivity to α-factor arrest [Fig. 5A, lanes 7 to 9]), and yet they had 10- to 20-fold more Cln than did ethanol-grown cells at Start (Fig. 5A and B). This shows that cells growing rapidly in sucrose require more Cln for Start than do cells growing slowly. Thus, the amount of Cln required for Start varies dramatically with the growth conditions (see Discussion).

The half-lives of CLN mRNAs and Cln proteins are similar at different proliferation rates.

Why are the amounts of Cln protein so low in slowly growing cells? It could be the ineluctable consequence of (i) a fixed, high rate of protein turnover in combination with (ii) a low overall rate of protein synthesis when cells are proliferating slowly. In other words, if Cln protein continues to turn over rapidly when overall protein synthesis is slow, then, other things being equal, there will necessarily be lower levels of Cln protein at low protein synthesis rates. To examine the plausibility of this hypothesis, we measured the half-lives of CLN mRNAs and Cln proteins in rapidly or slowly growing cells.

First, CLN mRNA half-lives were examined. In one experimental approach, we used the rpb1-1 mutation, which encodes a temperature-sensitive allele of RNA polymerase II such that transcription ceases abruptly upon a shift from 23 to 36°C (24). We grew cells in YEP-glucose or YEP-ethanol at 23°C, shifted them to 36°C, and measured CLN mRNA levels by Northern blotting at various times after the shift to the restrictive temperature. In all cases, the half-lives of the CLN1, CLN2, and CLN3 mRNAs expressed from their endogenous promoters were less than 5 min regardless of the carbon source or position in G₁ phase (data not shown).

Next, Cln protein half-lives were assayed. Once again, the rpb1-1 mutation and a shift from 23 to 36°C were used to shut off transcription and Western analysis was used to measure Cln levels as a function of time. In cells grown in glucose, the half-lives of Cln1 and Cln2 expressed from their endogenous promoters were 3 to 8 min (data not shown). Cln3 could not reliably be seen by Western analysis, so its half-life was not measured in these experiments. The experiment was then repeated with ethanol as a carbon source. Once again, Cln1 and Cln2 levels were much lower than in the glucose-grown cells (data not shown), but the half-lives were, again, about 3 to 8 min, the same as in the glucose-grown cells.

It is possible that the results were affected by the temperature shift. Furthermore, the very low levels of Cln1 and Cln2 in the ethanol-grown cells made accurate measurement difficult. Therefore, we did similar experiments with a promoter shutoff approach. A reg1 mutation was introduced into strains carrying GAL-CLN1-HA3 or GAL-CLN2-HA3. The reg1 mutation allows strong induction by galactose even in the presence of glucose. The two strains were grown in glucose or ethanol, and CLN1 or CLN2 was induced by the addition of galactose. After 1 h, cells were washed thoroughly to remove galactose and turn off the promoter and reinoculated into glucose or ethanol medium, as appropriate. Samples were taken at various times after the removal of galactose. In these experiments, the Cln1 and Cln2 proteins disappeared with a half-life of about 10 min in both the glucose-grown and ethanol-grown cells (Fig. 6). Thus, two different experimental approaches show that Cln1 and Cln2 are very unstable, even at low proliferation rates. Given that protein synthesis rates are low at low growth rates, and given that Cln turnover remains high, it is almost inevitable that Cln abundance will be low at low growth rates.

FIG. 6.

Cln protein half-lives are unaffected by changes in growth rate. Asynchronous cultures of strains with HA epitope-tagged Cln1 (GT141) or Cln2 (GT130) under the control of the GAL promoter were grown to mid-log phase in YEP plus either 2% glucose (A and C) or 3% ethanol (B and D). To induce the GAL promoter, galactose was added to 1% to each culture for 1 h. Cells were harvested by centrifugation and washed twice in YEP plus either 2% glucose (A and C) or 3% ethanol (B and D). Samples were taken after 0, 5, 10, 15, 20, and 45 min. The amount of Cln1 (A and B) or Cln2 (C and D) was assayed by Western blotting and normalized to β-tubulin (B-Tub) as a loading control. Lane Un shows an untagged control strain. (E) Quantification of Cln protein. Several exposures of these immunoblots were analyzed as described in Materials and Methods. Note that the exposure shown for panels A to D is largely saturated for the first few samples, and the quantitation of the early samples does not depend on this exposure. Cln protein was normalized to the β-tubulin loading control. The signal at zero time was set to 100%.

A system for controlled Cln expression.

To quantitatively evaluate whether a threshold amount of Cln is needed for Start, we developed a system to express controllable amounts of CLN mRNA. Strains containing integrated GAL-CLN constructs give high, nonphysiological levels of CLN expression when induced with 1% galactose. On the other hand, very low concentrations of galactose cannot be used because the cells metabolize galactose, leading to its rapid depletion. To create strains in which the GAL promoter could be induced by very low concentrations of galactose without such depletion, we deleted the promoters of the divergently transcribed GAL1 and GAL10 genes in a conditional CLN strain (cln1 cln2 cln3 GAL-CLN1), yielding gal1 gal10 strains. GAL1 encodes galactokinase, which is needed for the first step in galactose metabolism, so gal1 mutants have little or no ability to metabolize galactose (15). Thus, in a gal1 gal10 strain, galactose becomes a gratuitous inducer of the GAL promoter.

We first sought to determine if the gal1 gal10 cln1 cln2 cln3 GAL-CLN1 strain (BS111) could be used to express controlled levels of CLN1. To accomplish this, we grew the BS111 strain in YEP-1% raffinose-1% galactose to mid-log phase, washed it thoroughly with YEP-1% raffinose, resuspended it in YEP-1% raffinose, and then split the culture into seven aliquots. Different amounts of galactose (final percentages: 0, 0.001, 0.002, 0.003, 0.01, 0.03, and 0.1%) were added, and after 90 min cells were harvested and the level of CLN1 mRNA was assayed by Northern analysis. As shown in Fig. 7, there is a fairly direct relationship between the amount of galactose added and the level of expression of GAL-CLN1. In fact, between 0.001 and 0.03% galactose, the relationship is close to linear. At galactose concentrations greater than 0.03%, the response is partially saturated. Importantly, cells in 0.01% galactose expressed nearly wild-type levels of CLN1 mRNA (Fig. 7C) and Cln1 protein (Fig. 7D).

FIG. 7.

Experimental system for expressing controlled amounts of CLN1 mRNA. (A) Northern blot showing the amount of CLN1 mRNA induced by various amount of galactose in strain BS111 (gal1 gal10 cln1 cln2 cln3 GAL-CLN1). ACT1 was used as a loading and normalization control. (B) Quantitative data from part A are graphed. The x axis is a log scale. (C and D) Cells in 0.01% galactose express levels of CLN1 mRNA and Cln1 protein that are similar to wild-type (WT) levels. B-Tub, β-tubulin.

A graded response like the one shown in Fig. 7 could be explained if each cell in the population had a graded response to increasing galactose concentrations or if any one cell had an all-or-none response, but the proportion of responding cells increases with the galactose concentration. However, in this case, we can be certain that the second explanation is not correct because in this strain, viability depends on expression of GAL-CLN1. At very low concentrations of galactose, cells fail to bud and proliferate, and at slightly higher concentrations, nearly all of the cells bud and divide (Fig. 8). Thus, we conclude that the GAL promoter has a graded response to the galactose concentration. A graded response for the GAL promoter has also been seen by Biggar and Crabtree (1).

FIG. 8.

A critical Cln threshold for Start. Small unbudded G₁-phase cells of strain BS111 (cln1 cln2 cln3 GAL-CLN1 gal1 gal10) were obtained by centrifugal elutriation. The cells were divided into eight aliquots, and a different concentration of galactose was added to each aliquot: 1% (closed squares), 0.3% (closed diamonds), 0.1% (closed circles), 0.03% (closed triangles), 0.01% (open squares), 0.003% (open diamonds), 0.0001% (open circles), or 0% (open triangles). The percentage of budded cells, an indicator of progression past Start, was determined at various times. The percentage of budded cells was plotted versus cell size (A) or time (B). A galactose concentration of 0.03% or more allowed small cells to bud at sizes less than half that of the wild type (12 versus 27 fl). Cultures in 0.01% galactose (open squares) expressed roughly wild-type levels of CLN1 and budded at a size similar to that of the wild type (25 versus 27 fl). Cells with less than 0.01% galactose did not bud, even after very long incubations (data not shown).

We also measured the response to low concentrations of galactose as a function of time. After 6 h of induction, the amount of CLN1 mRNA had decreased to about half the level of the 90-min measurement (data not shown). There are many potential reasons for this loss of response with time; for instance, galactokinase may be expressed at low levels despite loss of its promoter. Alternatively, since measurements are being made per unit of total cellular RNA, this decrease in CLN1 mRNA could simply reflect the fact that the cells continue to grow in size, so the amount of normalizing, total RNA increases. In any case, because the loss of response was the same for 0.01 and 0.003% galactose over a 6-h period, it is reasonable to compare cellular responses to these levels of galactose over this time period.

Cells require a critical threshold amount of CLN expression for Start.

To determine if cells require a discrete level of CLN expression for Start, we used strain BS111 (cln1 cln2 cln3 GAL-CLN1 gal1 gal10). In this strain, CLN1, CLN2, and CLN3 have been deleted, so expression of GAL-CLN1 is required for Start and for viability. As discussed above, the strain carries a gal1 gal10 double mutation, allowing galactose to act as a gratuitous inducer at low concentrations.

Strain BS111 was grown in YEP-1% raffinose-1% galactose to mid-log phase, washed thoroughly with YEP-1% raffinose, and resuspended in YEP-1% raffinose to shut off GAL-CLN1. Subsequently, centrifugal elutriation was used to obtain small unbudded cells. These small unbudded cells (∼10 to 15 fl) were resuspended in fresh YEP-1% raffinose and split into eight fractions. In each fraction, a different amount of galactose was added (final percentages: 0, 0.001, 0.003, 0.01, 0.03, 0.1, 0.3, and 1%) and cells were incubated at 30°C. Samples were taken at regular intervals, and cell size and percent budding were measured. Results are shown in Fig. 8. In Fig. 8A, the budding data are plotted as a function of cell size; in Fig. 8B, the same budding data are plotted as a function of time.

There are three especially noteworthy points. First, it appears that there is a threshold requirement for CLN1, which is achieved at a galactose concentration of 0.01%. This gives a level of CLN1 mRNA that is very similar to the amount of CLN1 mRNA expressed from the natural CLN1 promoter (Fig. 7C). Moreover, these conditions result in physiological levels of Cln1 protein expression (Fig. 7D). A threefold lower level of galactose (0.003%) gives lower but detectable levels of CLN1 expression and yet cells fail to progress past Start. The cells in 0.003 and 0.001% galactose were also examined after 9 and 24 h, and there was still no budding (however, it should be noted that the amount of CLN1 mRNA in asynchronous BS111 cultures decreased twofold from 1.5 to 6 h, and even more after 9 and 24 h, after normalization to total RNA). On the other hand, threefold higher levels of galactose (0.03%) rapidly promoted progression past Start.

Second, when GAL-CLN1 expression is relatively high (0.03% galactose or more), budding occurs very rapidly and occurs at very small cell sizes (less than 20 fl). Under these conditions, a wild-type cell would not bud until it reached a size of about 27 fl. Budding at very small cell sizes suggests that Cln is either the main, or perhaps the only, limiting factor for Start.

Third, at the threshold level (0.01% galactose), the budding profiles versus size are remarkably similar to budding profiles in wild-type cells, even though the normal wave of SBF- and MBF-dependent transcription (which depends on Cln3) is presumably largely absent and even though the transcriptional regulation of CLN1 has an entirely different basis. Perhaps most strikingly, the cells grown in 0.01% galactose do not bud during the first 2 h (even though this is the period of maximal GAL-CLN1 expression) but then begin to bud in the third hour (when GAL-CLN1 expression is beginning to wane in asynchronous BS111 cultures!), a time when they achieved a size of about 25 fl, which is similar to the critical size for wild-type cells under these growth conditions. In other words, some mechanism is implementing relatively normal size control, despite the absence of CLN2 and CLN3 and the constitutive expression of CLN1.

Cln expression is size dependent.

Why do the cells grown in 0.01% galactose delay budding for 2 h, until they achieve a size of 25 fl? One possibility is that Cln1 works over time and that the cells require exposure to Cln1 for a substantial period of time before its work is done; i.e., the activity of Cln is integrated over time. A second possibility is that, as cells become larger, they become more sensitive to CLN1 mRNA (e.g., by synthesizing more protein, or, alternatively, by loss of an inhibitor of Cln function). To distinguish these possibilities, we designed an experiment to obtain unbudded, Cln-less G₁-phase cells of different sizes and then induce GAL-CLN1. To do this, strain BS111 was grown to mid-log phase in YEP-1% galactose-1% raffinose, washed with YEP-1% raffinose, and resuspended in YEP-1% raffinose to shut off GAL-CLN1. Subsequently, centrifugal elutriation was used to obtain small unbudded cells, exactly as for Fig. 8. These G₁-phase cells were resuspended in YEP-1% raffinose and split into three fractions, labeled small, medium, and large. The small cells (20 fl) were further split into three aliquots, and galactose was immediately added to a final concentration of 0, 0.003, or 0.01%. The medium cells were incubated in YEP-1% raffinose at 30°C until they grew to 30 fl, and then galactose was added to a final concentration of 0, 0.003, or 0.01%. The large cells were incubated in YEP-1% raffinose at 30°C until they grew to 40 fl, and then they were treated as described above. Thus, GAL-CLN1 was turned on in the small, medium, and large cells at sizes of 20, 30, and 40 fl, respectively.

None of the cells budded at any time in 0 or 0.003% galactose (data not shown). However, cells exposed to 0.01% galactose budded after a time that depended on their initial size (Fig. 9). The large cells were 50% budded after 45 min, the medium cells were 50% budded after 75 min, and the small cells were 50% budded after 150 min, by which time they had achieved a size of 31 fl. These results suggest that cell size, rather than the length of exposure to Cln1, is important.

FIG. 9.

Large cells are more sensitive to Cln than are small cells. Small unbudded G₁-phase cells of strain BS111 (cln1 cln2 cln3 GAL-CLN1 gal1 gal10) were obtained by centrifugal elutriation and split into three aliquots. To the one aliquot (closed triangles; 20-fl cells), 0.01% galactose was immediately added. The other two aliquots were incubated in galactose-free medium until they grew to a volume of 30 fl (closed squares) or 40 fl (closed circles), and only then was galactose added to 0.01%. The percentage of budded cells was assayed and is shown as a function of time from the addition of galactose (i.e., galactose was added at zero time). The small cells executed Start at ∼2.5 h after the addition of galactose, by which time they had grown to 31 fl. Large, medium, and small cells in just threefold less galactose (0.003%) failed to bud or execute Start at all, even after long incubations (data not shown).

It is possible that cell size affected the amount of CLN1 mRNA made in response to a given dose of galactose or, alternatively, that cell size affected the amount of Cln1 protein made in response to a given dose of mRNA. To investigate this, we examined the levels of CLN1 mRNA and Cln1 protein in the smallest fraction of cells as a function of time. During the time course (0 to 180 min), these cells (Fig. 9, closed triangles) grew to 38 fl. By measuring cell size at each time point, we found that the CLN1/ACT1 mRNA ratio was unaffected by cell size (Fig. 10A). In contrast, the level of Cln1 protein varied considerably with cell size (Fig. 10A and B). The amount of Cln1 protein expressed per unit of CLN1 mRNA increased dramatically as cells got larger (Fig. 10B). A 50% increase in cell size resulted in a fourfold increase in the amount of Cln1 protein expressed per unit of CLN1 mRNA; similarly, a 100% increase in cell size resulted in an eightfold increase in the amount Cln1 protein expressed per unit of CLN1 mRNA (Fig. 10B).

FIG. 10.

Cln protein expression is cell size dependent. (A) Small cells were harvested from the experiment described in Fig. 9 at 0, 60, 150, and 180 min (Fig. 9, closed triangles). Protein and RNA were isolated from each sample, and the amounts of Cln1 protein and CLN1 mRNA were determined by Western and Northern analyses. Northern analysis showed that CLN1 mRNA levels were largely unaffected by cell size (third blot from top). In contrast, Western analysis showed that Cln1 protein increased as a function of cell size (top blot). (B) The amount of Cln1 protein in each fraction of the small cells from panel A was first normalized to β-tubulin (B-Tub). Next, CLN1 mRNA levels were normalized to ACT1 mRNA levels. Shown here is a plot of the relative amounts of Cln1 protein (normalized protein levels divided by normalized mRNA levels) in fractions of increasing cell size. The time cells were incubated in 0.01% galactose is given for reference.

To investigate this observation further, we transformed strain BS111 with a plasmid containing an HA epitope-tagged CLN2 gene under the control of the Schizosaccharomyces pombe ADH promoter and largely lacking the natural 5′ untranslated region of CLN2. The S. pombe ADH promoter is constitutively expressed in S. cerevisiae and not regulated by cell size or cell cycle position (data not shown). Asynchronous cultures of BS111 transformed with this plasmid were grown to mid-log phase in YEP-2% glucose. Centrifugal elutriation was used to fractionate cells by size. Western analysis showed that larger cells contained significantly more Cln2 protein than did small cells (data not shown). Thus, we believe that the greater sensitivity of larger cells to CLN mRNA is because these larger cells maintain more Cln protein per unit of mRNA. We do not know whether this is a translational or posttranslational event.

DISCUSSION

Models and mechanisms linking Cln thresholds to Start.

At the outset of this work, we had in mind two simple models of how Cln activity might be related to Start. The first is the critical-threshold model (Fig. 11A), which asserts that Start occurs when the amount of Cln rises to some critical threshold. However, because of the instability of Cln mRNA and protein, it is not clear how slowly growing cells could ever attain the same levels of Cln as rapidly growing cells, and indeed, we have shown that they do not (Fig. 2, 4, and 5). We view the low level of Cln in slowly growing cells as the inevitable consequence of a combination of an unstable protein and a low rate of protein synthesis; the fact that the cells can nevertheless still pass through Start in a Cln-dependent way shows the existence of a compensatory mechanism. Thus, the critical-threshold model in its simplest form is incorrect (Fig. 4 and 5) since slowly growing and rapidly growing cells go through Start with very different amounts of Cln. Related results have previously been obtained by Heideman and coworkers (13, 25).

FIG. 11.

Working models for the Cln threshold. (A) Thresholds are shown in dotted lines, and theoretical Cln protein expression levels observed in rapidly and slowly growing cells are indicated. The critical-threshold model proposes that both slowly and rapidly growing cells must reach the same threshold for Cln activity to progress past Start. Results presented here argue against this model. (B) The integrated-activity model asserts that Cln proteins act over time during G₁ phase and that their cumulative effect, integrated over time (shown as the hatched areas under the curves), must reach some minimum to induce Start. (C) The variable-threshold model asserts that the critical threshold of Cln varies with the growth rate.

A second potential model is the integrated-activity model (Fig. 11B), which asserts that Cln proteins act over time during G₁ phase and that their cumulative effect, integrated over time, must reach some minimum to induce Start. In other words, a large amount of Cln could induce Start after a short G₁ phase or a small amount of Cln could induce Start after a longer G₁ phase. A similar idea has been proposed by Heideman and coworkers (13, 25). This model is also wrong in its simplest form; Fig. 8 shows that when Cln1 expression is reduced to one-third of its original level, Start never occurs, even after very long incubations. If the integrated-activity model were true, this reduction in Cln1 expression would simply expand G₁ phase by threefold. In addition, our results suggest that absolute cell size is important, and not just the time spent expressing Cln (Fig. 9 and 10; see also below).

Both models can be modified to fit the facts. With respect to the critical-threshold model, one could imagine that the Cln-Cdc28 complexes are opposed by some phosphatase or other activity and that the level of the phosphatase activity constitutes the threshold. If this phosphatase, like Cln1 and Cln2, had a half-life of about 10 min, then the level of the threshold would naturally go up and down with the growth rate, more or less paralleling the changes in Cln activity. This would be a variable-threshold model (Fig. 11C). With respect to the integrated-activity model, one could imagine that an inhibitor of Start (e.g., Sic1 or the hypothetical phosphatase) is expressed at a peak level in late M phase but also expressed at a lower level throughout the cell cycle. The inhibitor made in M phase could be phosphorylated (and then degraded?) by a large amount of Cln over a short time or by a small amount of Cln over a long time, but in either case, a critical minimum level of Cln would be needed to deal with the ongoing synthesis of the inhibitor.

Effect of cell size.

When GAL-CLN1 cells of different sizes were induced with galactose, they made the same amount of CLN1 mRNA (Fig. 10A). (By the same amount, we mean after normalization to ACT1 mRNA. In absolute terms, the large cells had more of the ACT1 and CLN1 mRNAs than did the small cells.) However, the larger cells went through Start much earlier than the smaller cells, showing, in some sense, a heightened sensitivity to CLN (Fig. 9). Larger cells also contained a larger amount of Cln protein than did the smaller cells (Fig. 10A and B) (again, after normalization to a control protein). Similar results were obtained with CLN2 expressed from the S. pombe ADH promoter. Thus, a larger cell size increases the relative amount of Cln protein by some posttranscriptional mechanism, and this may be essential for the accumulation of a critical amount of Cln.

In a wild-type cell, the ability of CLN3 to activate SBF and MBF transcription depends on cell size, even when the amount of CLN3 transcript remains relatively constant (32), and it has been suggested that Cln3 protein expression is also modulated by cell size (7). Thus, all three Cln proteins may increase in abundance with increased cell size, perhaps by a common posttranscriptional mechanism. The nature of this mechanism is unclear, but it could be increased translation. However, if increased translation is responsible, the effect does not seem to depend on the natural 5′ mRNA leaders of CLN1 and CLN2, which were absent in our experiments. Alternatively, it is possible that the Cln proteins are somehow selectively stabilized in larger G₁-phase cells.

Redundant cell size controllers.

Previous work clearly shows that CLN3 and its transcriptionally downstream effectors CLN1 and CLN2 are important for cell size control (10, 20, 35). Yet, in the experiments shown in Fig. 8 and 9, cell size control is relatively normal, despite the fact that CLN3 and CLN2 have been deleted, and CLN1 is being expressed at a constant level from the GAL promoter. Bck2 is partially redundant with Cln3 in providing size control, but it too is thought to work at least in part through the transcription of CLN1 and CLN2 (6, 9, 39), so it is not clear how Bck2 could play a role. That is, Fig. 8, 9, and 10 show cell size control at Start that is certainly independent of CLN3, and of the CLN1 and CLN2 promoters, and probably independent of BCK2. On the basis of Fig. 10, we imagine that some of this size control comes from the ability of large cells to generate relatively large amounts of protein from a given amount of CLN mRNA. This protein synthesis-based size control would thus be parallel to, and independent from, the CLN3/SBF/MBF pathway for controlling CLN transcription. Recent genome-wide screens have found many new genes that help control cell size, and many of these appear to work independently of CLN3 (16, 43). Strikingly, many of these size control genes are genes that affect ribosome biogenesis and so protein synthesis (e.g., SFP1, SCH9) (16, 43). One or more of these new size control-protein synthesis genes may be responsible for the redundant control mechanism we observe.

As a wild-type cell grows through G₁ phase, it transcribes increasingly larger amounts of CLN1, CLN2, and other SBF- and MBF-dependent genes (7). This increase in transcription with cell size is largely dependent on Cln3 and helps a cell pass through Start (37). At the same time, we now show that there is a novel Cln3-independent mechanism rendering Start more sensitive to a given amount of CLN1 or CLN2 transcript. That is, as cells grow they make more CLN1 and CLN2 transcripts in a CLN3-dependent way, but in addition, by a CLN3-independent mechanism, they generate a larger amount of Cln1 and Cln2 protein per transcript. These two effects may converge to convert moderate, gradual changes in size into a sharp, switch-like change in cell fate.